Dispersion Trading: Index vs Single-Stock Volatility and Correlation Trading

The Mathematical Foundation of Dispersion

Index volatility is related to single-stock volatilities through the portfolio variance formula. For an index composed of N stocks with weights w_i, volatilities σ_i, and pairwise correlations ρ_ij:

σ²(Index) = Σᵢ Σⱼ wᵢ wⱼ ρᵢⱼ σᵢ σⱼ

In simplified form for a large diversified index:
σ(Index) ≈ √ρ̄ × σ̄(Components)

Where ρ̄ is the average pairwise correlation and σ̄ is the average component volatility.

This equation has a critical implication: if you know the index volatility and the component volatilities, you can solve for the implied correlation. The options market prices both index options (giving you index implied vol) and single-stock options (giving you component implied vols). The difference between these two pieces of information — expressed as the implied correlation — is the central quantity in dispersion trading.

The Dispersion Premium: Why It Exists

Empirically, implied correlation (derived from the ratio of index IV to component IV) persistently exceeds realized correlation. In other words, the options market consistently prices index volatility too richly relative to what diversification mathematics would imply given the individual stock volatilities. This excess is called the dispersion premium, and it is the structural source of income for dispersion traders who sell index volatility and buy component volatility.

The dispersion premium exists for structural reasons:

• Demand asymmetry: Portfolio managers overwhelmingly want to buy index downside protection (SPX puts) as a hedge against their equity portfolios. This excess demand for index vol inflates index IV relative to what diversification would justify. The component vol market is less distorted because there is more balanced two-way flow in single-stock options.
• Correlation spikes in crises: During market crises, stock correlations spike toward 1 — all stocks fall together. The options market prices in a premium for this correlation spike risk, making index vol consistently more expensive than component vol on a correlation-adjusted basis. This risk premium is real, but it is over-priced — the market pays too much for it relative to how often the correlation spike materializes in the form of actual losses for dispersion sellers.
• Structured product supply: Banks sell volatility through structured products (equity-linked notes, variance swaps) and hedge by selling index variance and buying component variance — the dispersion trade. The persistent supply of index variance from structured product hedging keeps index IV rich.

Constructing the Dispersion Trade

A dispersion trade in its simplest form is:

• Short index variance (or sell index straddles/strangles)
• Long single-stock variance (or buy straddles/strangles) on the index components

The most professional implementation uses variance swaps — over-the-counter contracts that pay exactly the realized variance over the period minus the pre-agreed implied variance. Variance swaps eliminate all delta risk, leaving a pure volatility/variance exposure with no directional component.

For traders without access to variance swaps, the listed options equivalent is:

1. Sell SPX straddles or strangles (short index vol) — typically on the nearest monthly expiration, delta-hedged dynamically.
2. Buy straddles on top SPX components (long single-stock vol) — weighted by their index weight, delta-hedged dynamically. For SPX dispersion, top components include AAPL, MSFT, NVDA, AMZN, GOOGL, META, TSLA — the 10-15 largest positions represent a significant portion of index variance.

Weighting the Component Legs

The correct weighting for the component legs is not simply proportional to index weight. The vega-weighted approach is:

Component vega weight_i = (wᵢ² × σᵢ) / Σⱼ (wⱼ² × σⱼ)

Where wᵢ is the index weight and σᵢ is the component implied vol.

In practice, the exact weighting is complex and varies with the correlation structure. Most traders approximate by using index weights adjusted for relative vega, and accept some residual correlation exposure as the practical cost of trading the strategy in listed markets.

The Realized Correlation P&L Driver

The dispersion trade's P&L is driven primarily by the relationship between implied correlation and realized correlation over the life of the trade. The approximate P&L relationship:

Dispersion P&L ≈ (ρ_implied - ρ_realized) × Vega × σ_components

Profits when: realized correlation < implied correlation
Losses when: realized correlation > implied correlation

The trade is profitable when stocks move independently — idiosyncratic risk dominates, each company's stock moves based on its own fundamentals rather than macro factors driving all stocks in the same direction. It loses when stocks move together — during macro-driven crises where correlation spikes toward 1.

Risk Management: When Dispersion Fails

Dispersion trading has a well-documented asymmetric risk profile: frequent small gains from the dispersion premium, interrupted by occasional large losses when correlation spikes. Managing this risk requires:

Correlation Stress Testing

Every dispersion position should be stress-tested against the correlation spike scenario. The worst case for a dispersion trade is not just high realized correlation — it is the specific combination of high correlation AND high realized volatility that occurs during market crises. In 2008, realized correlation among S&P 500 stocks exceeded 0.8 while realized vol was above 50% — this combination produced severe losses for dispersion sellers who were not properly positioned for the tail risk.

Position Sizing for Correlation Risk

The correlation exposure of a dispersion book must be sized relative to the portfolio's ability to absorb the correlation spike scenario. A common rule used by vol desks: the maximum dispersion notional is sized so that the realized correlation spike to 0.9 (the historical 95th percentile) would produce a loss of no more than X% of the overall portfolio. The value of X varies by institution but typically ranges from 2-5% of portfolio NAV per single expiration cycle.

Entry Timing

The implied correlation level at trade entry is the primary determinant of expected return. Entering when implied correlation is at 0.5 (VIX in the 12-16 range, low fear) offers a much smaller premium than entering when implied correlation has spiked to 0.8+ during a crisis. Counterintuitively, the best time to enter a dispersion trade (short correlation) is often during or after a volatility event, when implied correlation is elevated and the trade offers maximum risk premium.

Dispersion Signals for Non-Dispersion Traders

Even traders who do not execute dispersion trades directly can extract valuable signals from dispersion market data:

• Implied correlation as a fear gauge: Rising implied correlation means the options market is pricing in more macro-driven synchronized moves. This is typically a risk-off signal — when traders expect stocks to move together, they are expecting macro fear rather than idiosyncratic opportunity.
• Dispersion premium as a VIX context indicator: A high dispersion premium (implied correlation well above realized) means index vol is particularly rich relative to components. This is a context signal for VIX-short strategies — if you are planning to sell index vol, a wide dispersion premium tells you that index vol is expensive in relative terms, confirming the short vol thesis.
• Dispersion collapse as a warning: When the dispersion premium collapses (implied correlation approaches realized correlation from above), it often signals that the structural bid for index protection is falling — perhaps because institutional hedgers are reducing their portfolios. Falling dispersion premium can be an early indicator of broader institutional risk reduction.

CrossVol tracks implied correlation in real-time — derived from SPX/ES options and major component options — providing the dispersion premium calculation alongside the GEX and VPIN data that contextualizes the current correlation regime and its likely direction.